Resolving to Towards a Hanging-Rail Register: A regression-discontinuity study of the eighteen-point print-night cutoff